Sudden Infant Death Syndrome and Maternal Smoking Michael H. Malloy, MD, MS, Howard J. Hoffmnan, AL4, and Donald R Peterson, MD, MPH
Inrodudion A number of socioeconomic and demographic factors are associated with an increased risk of sudden infant death syndrome (SIDS), but few readily amenable exposure factors have been identified.1 Reports of an association between maternal cigarette smoking during pregnancy and the subsequent occurrence of SIDS, however, do offer the possibility that reducing maternal smoking could result in a decline in the SIDS rate.2-4 Whether the association between SIDS and a history of maternal smoking during pregnancy is biological in nature or a proxy for maternal behavior is not clear. Haglund and Cnattingius reported that infants born to women who smoke during pregnancy die earlier with SIDS than infants of nonsmokers and that a doseresponse association exists between maternal smoking and the risk of SIDS.5 Their report supports the plausibility of a biological mechanism. We attempted to confirm their observations in two large US data sets.
Methods The Missouri data set is a linked birth and death certificate file for the period 1980 to 1985. There were 684 SIDS deaths in Missouri during this period. Complete data were available on 636 of these cases and 425 326 live births, an overall rate of 1.6/1000 live births (1.3 non-Black vs 3.1 Black). For this analysis, we excluded infants of gestational age less than 24 weeks or greater than 45 weeks and excluded cases of SIDS in which the age of death was less than 7 days. Seventy-six percent of the SIDS deaths reported between 1980 and 1985 were autopsied, but no other special review process was carried out to confirm the validity of the assignment of the diagnosis of SIDS. Additional information obtained from the birth certificate included race, maternal age, parity, maternal education, and marital status. Maternal smoking information was obtained from the birth certificate and was reported in terms of packs of cigarettes smoked per day.
The National Institute of Child Health and Human Development SIDS Cooperative Epidemiological Study was a case-control study conducted between October 1978 and December 1979 that accrued cases of SIDS across six geographic regions.6 The SIDS rate for this study was 1.75/1000 live births (1.10 nonBlack vs 3.65 Black). All cases of SIDS were autopsied, and a panel of three expert pathologists reviewed each of the cases to determine the reliability of the diagnosis. Living controls (Control A) were selected from regional birth certificate files to match regional cases on the age of death. A second living control (Control B) was matched to the SIDS case within each region on age, race, and birthweight (c 2500 g). There were 757 SIDS cases and 1514 controls available for this analysis. Mothers of SIDS cases were interviewed 2 to 5 weeks after the child died. Mothers of cases and controls were asked during the interview how many cigarettes per day they smoked during the first, second, and third trimesters. For this analysis, information on smoking in the first trimester was used. Analyses were carried out to determine whether the SIDS infants' age of death was less when the mother smoked than when the mother did not smoke by using general chi-square tests of association. The relationship between the quantity smoked and infant mortality was analyzed by logistic regression.7 Two-way interactions between smoking and covariates in the logistic models were examMichael H. Malloy and Howard J. Hoffinan are with the Division of Epidemiology, Statistics and Prevention Research, National Institute of Child Health and Human Development, Bethesda, Md. Michael H. Malloy is also with the Department of Pediatrics, University of Texas Medical Branch, Galveston. Donald R. Peterson is with the Departnent of Biostatistics, University of Washington, Seattle. Requests for reprints should be sent to Michael H. Malloy, MD, MS, Department of Pediatrics, Rte E-26, University of Texas Medical Branch, Galveston, TX 77550. This paper was submitted to the Journal October 8, 1991, and accepted with revisions May 26, 1992.
October 1992, Vol. 82, No. 10
Public Health Briefs
ined, but none were significant. Analyses done with SAS.8
were
Percent Dead
Rests In the Missouri population and in the Cooperative Study, the distribution of the age of SIDS death for smokers and nonsmokers appears quite similar (Figures 1 and 2). There were no statistically significant associations by general chisquare analysis. The median age of death for both nonsmokers and smokers was 11 weeks. This contrasts with the observation from Sweden of a median age of SIDS death among nonsmokers of 10.4 weeks compared with 8 weeks among smokers.5 The dose-response relationship between smoking and SIDS was examined by logistic regression (Table 1). The odds ratios were adjusted by logistic regression for race, parity, age of the mother, maternal education, marital status, and sex of the infant. For Missouri, the adjusted odds ratios were 1.98 for light smokers compared with nonsmokers and 2.86 for heavy smokers compared with nonsmokers. The odds ratio for heavy smokers compared with light smokers was 1.41 with a 95% confidence interval (CI) that did exclude 1.0. For the Cooperative Study, the adjusted odds ratios, with the reference group being nonsmokers and using Control A (the controls matched on age alone), were 2.49, 3.24, and 3.05 for smokers of less than 10, 10 through 19, and 20 or more cigarettes per day, respectively. For Control B (the controls matched on age, birthweight, and race), the odds ratios were 2.68, 2.65, and 2.60,
Chi-square P f.O2S
25 200h
3.59
.!
WAR-QMMIw.Amra. OCmaLkara
16 10
-
5 BI
41
1
2
3
4
6
)W7
6
Age of Death (months) FIGURE 1Pere distridbuton of the age of death (in months) by maternal smoking status for sudden infant death syndrome deaths occurring in Missouri for the years 1980 to 1985. IR
36
Percent Dead
30
Chi-square i9.35
0.23
26
20 _P_
16...
_..~_NnSm-okers
.
10 *L
respectively. 6 iscussion
We were unable to confirm the Swedish observation of a relationship between the age of death and a history of maternal smoking during pregnancy. In the Missouri population, we did observe a relationship between the quantity smoked and the risk for SIDS. Data from the Cooperative Study, however, failed to support a dose-response relationship. Differences in results between the two populations may be due to recall bias. In the Cooperative Study women reported smoking after the death of the infant, while in the Missouri population smoking information was reported on the birth
October 1992, Vol. 82, No. 10
I
0
'1
1
2
3
4
5
6
7
Age of Death (months) FIGURE 2-Percente dlsbuion of the age of death (n months) by maermal smokIng statsforsudden infantdeath syndrome (SIDS) deathsoccurring Inthe SIDS CooperNal inti of Child Health and Human Devel atve Study.
certificate prior to the death of the infant. However, the direction of this potential bias is uncertain. Confirmation of the Swedish report of an earlier age of death among infants of smokers would support the hypothesis that the relationship between maternal
and SIDS is biologic. However, unable to demonstrate this relationshipwith two large US data sets. Nevertheless, further research into biologic mechanisms and life-style factors by which maternal smoking might increase the risk of SIDS should be pursued. smoking
we were
American Journal of Public Health 1381
Public Health Briefs
knowledge the principal investigators and their staffs, including the scientific advisors and consultants, of the National Institute of Child Health and Human Development SIDS Cooperative Epidemiological Study.
References 1. Damus K, Pakter J, Drongrad E, Standfast SJ, Hoffman HI. Postnatal medical and epidemiological risk factors for the sudden infant death syndrome. In: Harper RM, Hoffman HJ, eds. Sudden Infant Death Syndmne: Risk Factors and Basic Mechanisns. New York, NY: PMA Publishing Corp; 1988:187-201. 2. Naeye RL, Ladis B, Drage JS. Sudden infant death syndrome: a prospective study.
Is~~~~~I Mw~~~I
=~~~~~~~~~~~~~~~~~~~~~~~~~~~~r,0
Acknowledgments
ADC. 1976;130:1207-1210. 3. Bergman AB, Wiesner LA. Relationship of passive cgarette smoking to sudden infant death syndrome. Pediatrics. 1976;58:665668. 4. Malloy MH, Kleinman JC, Land GH, Schramm WF. The association of maternal smokingwith age and cause of infant death. Am JEpidemioL 1988;128:46-55. 5. Haglund B, Cnattingius S. Cigarette smoking as a risk factor for sudden infant death syndrome: a population-based study. Am J Public Healt 1990;80:29-32. 6. Hoffman ]H, Damus K, Hilhman L, Krongrad E. Riskfactors for SIDS: results of the National Institute of Child Health and Human Development SIDS Cooperative Epidemiological Study. Ann NY Acad Sci.
An abstract of this research was presented at the April 1991 joint annual meeting ofthe American Pediatric Society and the Society for Pediatrics Research in New Orleans, La. We would like to thank Garland Land and his staff at the Missouri Department of Health for makingvital statistics data available for this research study. Also, we would like to ac-
7. Kleinbaum DG, Kupper LL, Morgenstern H. Epidemiologic Research Principles and Quantitate Methds. Belmont, Calif: Lifetime Learning Publications; 1982. 8. SAS Institute Inc. &4S User's Guide: Statisfics, Version 5 EditioL Cary, NC: SAS Institute Inc; 1985.
2*~~~~~~~~.
Public health programs that encourage smoking reduction for pregnant women offer the possibility of effecting a more immediate decline in SIDS. It remains to be determined, however, whether the benefits of public health programs that encourage smoking reduction will include a decline in SIDS occurrence. E
k
1988;533:13-30.
Ca-l for HI*Niau Papers on Pkdhic Health Nwsing Ecuse in 1993 public health nurses celebrate their coentenil historians ofpublichealthnutSmigareiwltedto mt pla pu by the artic for review for cen jona PFkbic Healh Nw . i s for authors are prined in the back coer of the joual. I es are p l
1382 American Journal of Public Health
in March, June, Septmber, and D m and debes are a to 6 months tion. For a please cal Karen Buhler-Wilkerson, hitorical editor, Pkuuc
Healh Nw*nw
at (215) 898-4725.
October 1992, Vol. 82, No. 10
Inrodudion A number of socioeconomic and demographic factors are associated with an increased risk of sudden infant death syndrome (SIDS), but few readily amenable exposure factors have been identified.1 Reports of an association between maternal cigarette smoking during pregnancy and the subsequent occurrence of SIDS, however, do offer the possibility that reducing maternal smoking could result in a decline in the SIDS rate.2-4 Whether the association between SIDS and a history of maternal smoking during pregnancy is biological in nature or a proxy for maternal behavior is not clear. Haglund and Cnattingius reported that infants born to women who smoke during pregnancy die earlier with SIDS than infants of nonsmokers and that a doseresponse association exists between maternal smoking and the risk of SIDS.5 Their report supports the plausibility of a biological mechanism. We attempted to confirm their observations in two large US data sets.
Methods The Missouri data set is a linked birth and death certificate file for the period 1980 to 1985. There were 684 SIDS deaths in Missouri during this period. Complete data were available on 636 of these cases and 425 326 live births, an overall rate of 1.6/1000 live births (1.3 non-Black vs 3.1 Black). For this analysis, we excluded infants of gestational age less than 24 weeks or greater than 45 weeks and excluded cases of SIDS in which the age of death was less than 7 days. Seventy-six percent of the SIDS deaths reported between 1980 and 1985 were autopsied, but no other special review process was carried out to confirm the validity of the assignment of the diagnosis of SIDS. Additional information obtained from the birth certificate included race, maternal age, parity, maternal education, and marital status. Maternal smoking information was obtained from the birth certificate and was reported in terms of packs of cigarettes smoked per day.
The National Institute of Child Health and Human Development SIDS Cooperative Epidemiological Study was a case-control study conducted between October 1978 and December 1979 that accrued cases of SIDS across six geographic regions.6 The SIDS rate for this study was 1.75/1000 live births (1.10 nonBlack vs 3.65 Black). All cases of SIDS were autopsied, and a panel of three expert pathologists reviewed each of the cases to determine the reliability of the diagnosis. Living controls (Control A) were selected from regional birth certificate files to match regional cases on the age of death. A second living control (Control B) was matched to the SIDS case within each region on age, race, and birthweight (c 2500 g). There were 757 SIDS cases and 1514 controls available for this analysis. Mothers of SIDS cases were interviewed 2 to 5 weeks after the child died. Mothers of cases and controls were asked during the interview how many cigarettes per day they smoked during the first, second, and third trimesters. For this analysis, information on smoking in the first trimester was used. Analyses were carried out to determine whether the SIDS infants' age of death was less when the mother smoked than when the mother did not smoke by using general chi-square tests of association. The relationship between the quantity smoked and infant mortality was analyzed by logistic regression.7 Two-way interactions between smoking and covariates in the logistic models were examMichael H. Malloy and Howard J. Hoffinan are with the Division of Epidemiology, Statistics and Prevention Research, National Institute of Child Health and Human Development, Bethesda, Md. Michael H. Malloy is also with the Department of Pediatrics, University of Texas Medical Branch, Galveston. Donald R. Peterson is with the Departnent of Biostatistics, University of Washington, Seattle. Requests for reprints should be sent to Michael H. Malloy, MD, MS, Department of Pediatrics, Rte E-26, University of Texas Medical Branch, Galveston, TX 77550. This paper was submitted to the Journal October 8, 1991, and accepted with revisions May 26, 1992.
October 1992, Vol. 82, No. 10
Public Health Briefs
ined, but none were significant. Analyses done with SAS.8
were
Percent Dead
Rests In the Missouri population and in the Cooperative Study, the distribution of the age of SIDS death for smokers and nonsmokers appears quite similar (Figures 1 and 2). There were no statistically significant associations by general chisquare analysis. The median age of death for both nonsmokers and smokers was 11 weeks. This contrasts with the observation from Sweden of a median age of SIDS death among nonsmokers of 10.4 weeks compared with 8 weeks among smokers.5 The dose-response relationship between smoking and SIDS was examined by logistic regression (Table 1). The odds ratios were adjusted by logistic regression for race, parity, age of the mother, maternal education, marital status, and sex of the infant. For Missouri, the adjusted odds ratios were 1.98 for light smokers compared with nonsmokers and 2.86 for heavy smokers compared with nonsmokers. The odds ratio for heavy smokers compared with light smokers was 1.41 with a 95% confidence interval (CI) that did exclude 1.0. For the Cooperative Study, the adjusted odds ratios, with the reference group being nonsmokers and using Control A (the controls matched on age alone), were 2.49, 3.24, and 3.05 for smokers of less than 10, 10 through 19, and 20 or more cigarettes per day, respectively. For Control B (the controls matched on age, birthweight, and race), the odds ratios were 2.68, 2.65, and 2.60,
Chi-square P f.O2S
25 200h
3.59
.!
WAR-QMMIw.Amra. OCmaLkara
16 10
-
5 BI
41
1
2
3
4
6
)W7
6
Age of Death (months) FIGURE 1Pere distridbuton of the age of death (in months) by maternal smoking status for sudden infant death syndrome deaths occurring in Missouri for the years 1980 to 1985. IR
36
Percent Dead
30
Chi-square i9.35
0.23
26
20 _P_
16...
_..~_NnSm-okers
.
10 *L
respectively. 6 iscussion
We were unable to confirm the Swedish observation of a relationship between the age of death and a history of maternal smoking during pregnancy. In the Missouri population, we did observe a relationship between the quantity smoked and the risk for SIDS. Data from the Cooperative Study, however, failed to support a dose-response relationship. Differences in results between the two populations may be due to recall bias. In the Cooperative Study women reported smoking after the death of the infant, while in the Missouri population smoking information was reported on the birth
October 1992, Vol. 82, No. 10
I
0
'1
1
2
3
4
5
6
7
Age of Death (months) FIGURE 2-Percente dlsbuion of the age of death (n months) by maermal smokIng statsforsudden infantdeath syndrome (SIDS) deathsoccurring Inthe SIDS CooperNal inti of Child Health and Human Devel atve Study.
certificate prior to the death of the infant. However, the direction of this potential bias is uncertain. Confirmation of the Swedish report of an earlier age of death among infants of smokers would support the hypothesis that the relationship between maternal
and SIDS is biologic. However, unable to demonstrate this relationshipwith two large US data sets. Nevertheless, further research into biologic mechanisms and life-style factors by which maternal smoking might increase the risk of SIDS should be pursued. smoking
we were
American Journal of Public Health 1381
Public Health Briefs
knowledge the principal investigators and their staffs, including the scientific advisors and consultants, of the National Institute of Child Health and Human Development SIDS Cooperative Epidemiological Study.
References 1. Damus K, Pakter J, Drongrad E, Standfast SJ, Hoffman HI. Postnatal medical and epidemiological risk factors for the sudden infant death syndrome. In: Harper RM, Hoffman HJ, eds. Sudden Infant Death Syndmne: Risk Factors and Basic Mechanisns. New York, NY: PMA Publishing Corp; 1988:187-201. 2. Naeye RL, Ladis B, Drage JS. Sudden infant death syndrome: a prospective study.
Is~~~~~I Mw~~~I
=~~~~~~~~~~~~~~~~~~~~~~~~~~~~r,0
Acknowledgments
ADC. 1976;130:1207-1210. 3. Bergman AB, Wiesner LA. Relationship of passive cgarette smoking to sudden infant death syndrome. Pediatrics. 1976;58:665668. 4. Malloy MH, Kleinman JC, Land GH, Schramm WF. The association of maternal smokingwith age and cause of infant death. Am JEpidemioL 1988;128:46-55. 5. Haglund B, Cnattingius S. Cigarette smoking as a risk factor for sudden infant death syndrome: a population-based study. Am J Public Healt 1990;80:29-32. 6. Hoffman ]H, Damus K, Hilhman L, Krongrad E. Riskfactors for SIDS: results of the National Institute of Child Health and Human Development SIDS Cooperative Epidemiological Study. Ann NY Acad Sci.
An abstract of this research was presented at the April 1991 joint annual meeting ofthe American Pediatric Society and the Society for Pediatrics Research in New Orleans, La. We would like to thank Garland Land and his staff at the Missouri Department of Health for makingvital statistics data available for this research study. Also, we would like to ac-
7. Kleinbaum DG, Kupper LL, Morgenstern H. Epidemiologic Research Principles and Quantitate Methds. Belmont, Calif: Lifetime Learning Publications; 1982. 8. SAS Institute Inc. &4S User's Guide: Statisfics, Version 5 EditioL Cary, NC: SAS Institute Inc; 1985.
2*~~~~~~~~.
Public health programs that encourage smoking reduction for pregnant women offer the possibility of effecting a more immediate decline in SIDS. It remains to be determined, however, whether the benefits of public health programs that encourage smoking reduction will include a decline in SIDS occurrence. E
k
1988;533:13-30.
Ca-l for HI*Niau Papers on Pkdhic Health Nwsing Ecuse in 1993 public health nurses celebrate their coentenil historians ofpublichealthnutSmigareiwltedto mt pla pu by the artic for review for cen jona PFkbic Healh Nw . i s for authors are prined in the back coer of the joual. I es are p l
1382 American Journal of Public Health
in March, June, Septmber, and D m and debes are a to 6 months tion. For a please cal Karen Buhler-Wilkerson, hitorical editor, Pkuuc
Healh Nw*nw
at (215) 898-4725.
October 1992, Vol. 82, No. 10
